Jens Lange, Jonas Dalege, Denny Borsboom, Gerben A. van Kleef & Agneta H. Fischer (2020), Perspectives on Psychological Science 15(2):444–468 — DOI-10.1177/1745691619895057
Key takeaways
- Three properties (pp. 445–448). From affect-program theories, distinct emotions. From constructionist theories, variation between and within persons. From appraisal theories, causal relations among a multitude of components. A psychometric model is a construct theory, so a model with all three would help unify the theories.
- Latent-variable models fail (pp. 448–455). A reflective model makes the emotion a common cause separate from its components, and implies local independence, which evidence contradicts. A formative model makes the emotion an index of its components, which leaves the components’ relations uninformative.
- The network model (pp. 455–458). “For emotions, this would imply that the network of causal relationships between emotion components represents the emotion” (p. 456); “the network constitutes the emotion” (p. 456). A distinct emotion is a small-world community of clusters, weakly emergent: “a stable system can emerge independently of which emotion component was activated first or whether multiple components were activated simultaneously (i.e., the emergent state is autonomous). These systems of interactions then force each emotion component into a specific state (i.e., weak downward causation)” (p. 457).
- Prediction: duration (p. 460). “As the connections are strong, the components continuously reactivate each other, sustaining the activity of the network. Thereby, the duration of the emotion is extended, and it becomes more difficult to switch back to a deactivated state.” This is hysteresis, and “In less strongly connected networks, hysteresis does not occur.” Appraised relevance may strengthen the connections.
- Limits they name (pp. 460–461). Edge weights say nothing about how far components are endorsed. Cross-sectional networks can hide different persons’ structures. Cross-sectional, contemporaneous and temporal networks can disagree in sign. Estimation is exploratory.
Standing in the record
Filed on 2026-10-23 for the record’s proposal that hedonic valence is an ordering’s potential in the negotiation among perspectives (CLAIM-440, TERM-076), and for the measurement question QUESTION-268. It is the emotion-domain counterpart of Borsboom’s network theory of mental disorders (LIT-547, THEORY-058), which the record holds. Lange et al. give persistence a mechanism, self-sustaining activation in a strongly connected network with hysteresis. But what persists is a pattern of components, not one perspective’s ranking prevailing over others, and the model has no sign. The bearing is set out in NOTE-1098.
No instruction for machine-learning practice; nothing here belongs in the anthology.